Estrogen Receptor B Polymorphism Is Associated with Prostate

Cancer Prevention

Estrogen Receptor B Polymorphism Is Associated with Prostate Cancer Risk Camilla Thellenberg-Karlsson,1 Sara Lindstro¨m,1 Beatrice Malmer,1 Fredrik Wiklund,1 Katarina Augustsson-Ba¨lter,3 Hans-Olov Adami,3 Par Stattin,2 Maria Nilsson,4 Karin Dahlman-Wright,4 Jan-A˚ke Gustafsson,4 and Henrik Gro¨nberg1

Abstract

Purpose: After cloning of the second estrogen receptor, estrogen receptor h (ERh) in 1996, increasing evidence of its importance in prostate cancer development has been obtained. ERh is thought to exert an antiproliferative and proapoptotic effect. We examined whether sequence variants in the ERh gene are associated with prostate cancer risk. Experimental Design: We conducted a large population-based case-control study (CAncer Prostate in Sweden, CAPS) consisting of 1,415 incident cases of prostate cancer and 801 controls.We evaluated 28 single nucleotide polymorphisms (SNP) spanning the entire ERh gene from the promoter to the 3V-untranslated region in 94 subjects of the control group. From this, we constructed gene-specific haplotypes and selected four haplotype-tagging SNPs (htSNP: rs2987983, rs1887994, rs1256040, and rs1256062). These four htSNPs were then genotyped in the total study population of 2,216 subjects. Results: There was a statistically significant difference in allele frequency between cases and controls for one of the typed htSNPs (rs2987983), 27% in cases and 24% in controls (P = 0.03). Unconditional logistics regression showed an odds ratio of 1.22 (95% confidence interval, 1.02-1.46) for men carrying the variant allele TC or CC versus the wild-type TT, and an odds ratio of 1.33 (95% confidence interval, 1.08-1.64) for localized cancer. No association of prostate cancer risk with any of the other SNPs or with any haplotypes were seen. Conclusion:We found an association with a SNP located in the promoter region of the ERh gene and risk of developing prostate cancer. The biological significance of this finding is unclear, but it supports the hypothesis that sequence variation in the promoter region of ERh is of importance for risk of prostate cancer.

Prostate cancer is a large global health problem with as many as half a million new cases each year (1). Genetic susceptibility is of major importance in the etiology of prostate cancer and may account for as much as 40% of all cases (2). A recent segregation analysis (3) suggests that multiple low-penetrant genes account for a major part of the genetic susceptibility of prostate cancer and that only a small fraction can be explained by dominant inheritance of highly penetrant genes.

Authors’ Affiliations: Departments of 1Radiation Sciences/Oncology, and 2 Surgical and Perioperative Sciences, University of Umea˚, Umea˚, 3Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, and 4 Department of Medical Nutrition and Biosciences, Karolinska Institutet, Novum, Huddinge, Sweden Received 2/4/05; revised 11/8/05; accepted 12/15/05. Grant support: Lion’s Cancer Research Foundation (Umea˚, Sweden), Swedish Cancer Society, and Spear grant fromthe Umea˚ University Hospital (Umea˚, Sweden). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section1734 solely to indicate this fact. Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). Requests for reprints: Henrik Gro¨nberg, Department of Radiation Sciences/ Oncology, Umea˚ University, S-901 87 Umea˚, Sweden. Phone: 46-90-785-1982; Fax: 46-90-127-464; E-mail: Henrik.Gronberg@ oc.umu.se. F 2006 American Association for Cancer Research. doi:10.1158/1078-0432.CCR-05-0269

Clin Cancer Res 2006;12(6) March 15, 2006

The estrogen receptor h (ERh) gene is highly expressed in the prostate epithelium, suggesting a direct effect of estrogen on the prostate (4, 5). Deletion of ERh in mice lead to hyperplasia in the ventral prostate, indicating that ERh has an antiproliferative role in this tissue (6), a notion that was later confirmed (7). A number of studies on the expression pattern of ERh in both normal and malignant prostate tissue have been done. In most studies of cancer, expression of ERh diminishes with increasing Gleason score but may reappear in metastatic lesions (8 – 11). Epigenetic regulation, for example, methylation of the promoter region seems to cause down-regulation of downstream genes. It also seems to be a reversible event in tumor progression because methylation is more common in high-grade compared with low-grade prostate cancer and normal prostate tissue, and then decreases again in metastatic lesions compared with localized high-grade tumors (9, 12, 13). Intense staining with Ki-67 in the prostate of ERh knockout mice, together with high expression of the androgen receptor, suggests the regulatory role of ERh in the androgen receptor. A proposed ligand to ERh, 5a-androstane 3h, 17h-diol (3hAdiol) inhibits the growth of prostate epithelium in wild-type mice but not in ERh knockout mice, via down-regulation of androgen receptor (6, 14). Other studies of prostate cancer cell lines have shown the antiproliferative and anti-invasion properties of ERh as reintroduction of the gene inhibits growth and invasion and triggers apoptosis (15).

1936

www.aacrjournals.org

Downloaded from clincancerres.aacrjournals.org on September 2, 2018. © 2006 American Association for Cancer Research.

ERb Polymorphisms Associated with Prostate Cancer Risk

Our hypothesis is that genetic variation in the ERh gene might alter the expression of the gene and thus affect the risk of prostate cancer. We tested this hypothesis in a large populationbased epidemiologic study.

Patients and Methods Briefly, the cases and controls studied came from a large-scale, population-based case-control study (CAncer Prostate in Sweden, CAPS). A detailed description of the CAPS study is presented elsewhere (16). The study population consisted of 1,415 patients with prostate cancer and 801 control subjects. The case participants were recruited from four of the six regional cancer registries that cover the entire population of Sweden. Each of these registries serves one health care region (Northern, Central, Stockholm, and Southeastern) and altogether encompasses f6 million inhabitants (67% of Sweden’s population). The cases were linked to the National Prostate Cancer Registry and clinical information such as Gleason sum, PSA level at the time of diagnosis, tumor-node-metastasis stage, means of diagnosis and primary treatment were obtained for 95.3% of the cases. The cases were thereafter classified as either localized (n = 772); (T1-2 and N0/NX and M0/MX and grades 1-2/Gleason sum 2-7, and PSA 100). For cases and controls who reported at least one family member with prostate cancer, a more detailed family history of prostate cancer was obtained through additional questionnaires and record linkage to the Swedish Cancer Registry or through medical records. After retrieving these supplementary data, a total of 177 cases were classified as familial prostate cancer. Control subjects were randomly selected from the continuously updated Swedish Population Registry, frequency-matched according to the expected age distribution (within 5 years) and geographic origin of the cases. Mean age (age at diagnosis for case patients and age at inclusion for control subjects) for the cases and controls were 66.6 and 67.9 years, respectively. The study was approved by the Ethical Committees at the two participating academic institutions, Umea˚ University and Karolinska Institutet. Written informed consent was obtained from each subject. Selection of ER-b single nucleotide polymorphisms. To make a thorough evaluation of sequence variation in the ERh gene, we used a haplotype-tagging single nucleotide polymorphisms (htSNP) method. The ER-h gene is located on chromosome 14 q23.2, and is f61.2 kb including eight exons, together with two untranslated first exons, ON and OK (Fig. 1). Five isoforms are presently known (17) and even more mRNA splice variants of unknown importance seem to exist (18). We conducted a search for known SNPs in the data bases http:// snpper.chip.org/ and http://www.ncbi.nih.gov/SNP/ and selected a subset of SNPs from the promoter region (15 kb), introns, exons, and 3V-untranslated region (UTR) covering a total length of 68.5 kb. There are three SNPs in the coding regions, all synonymous. At the time of selection, not many SNPs in ERh were validated and even fewer had frequency data, so the main criteria for selection were that the SNPs were evenly spread throughout the gene. In total, 37 SNPs were chosen with a mean distance between SNPs of 1,800 bp (Table 1). The SNP genotyping assays were designed by using the Assay-by-Design and Assay-on-Demand service (Applied Biosystems, Foster City, CA). The 37 SNPs were then genotyped in 94 randomly selected control subjects using a 5V-nuclease TaqMan assay together with fluorescently labeled Minor Groove Binders probes. All reactions were done in a 25 AL volume consisting of 10 ng of genomic DNA, 900 nmol/L of each primer, 200 nmol/L of each probe and 12.5 AL of TaqMan universal master mix. PCR cycling conditions were: 50jC for 2 minutes, 95jC for 10 minutes followed by 40 cycles of 92jC for 15 seconds, and 60jC for 1 minute. The samples were analyzed on an ABI 7700 sequence detection system. Five of the genotyped SNPs were monomorphic in the 94 randomly selected

www.aacrjournals.org

subjects and therefore excluded from further analysis, as were 10 SNPs with assay failure. From the remaining 22 SNPs, haplotypes were estimated using a Markov Chain Monte Carlo approach as implemented in the PHASE software package (http://www.stats.ox.ac.uk/ mathgen/software.html). Four SNPs (rs2987983, rs1887994, rs1256040, and rs1256062), which captured 99.6% of the haplotype variation among the 94 controls, were selected as htSNPs using the htSNP2 package (http:// www.gene.cimr.cam.ac.uk/clayton/software/stata) for the STATA software. These four htSNPs were genotyped under the same conditions and with the same equipment as described above for all 1,415 cases and 801 control subjects. For all of the htSNPs, three htSNPs in each group of homozygous and heterozygous subjects were sequenced to produce internal controls. The primer sequences used to amplify the target DNA are available on request. PCR conditions were as follows: deoxynucleotide triphosphate 200 Amol/L, 10 PCR buffer, 3 mmol/L MgCl2, 0.5 Amol/L primer, 1 unit Taq Gold enzyme, 50 ng of DNA, cycling at 94jC for 20 seconds, 50jC for 30 seconds, 72jC for 30 seconds, repeated 35 cycles. For optimizing the PCR for SNP rs1256040, the following changes were made: increased MgCl2 to 4.5 mmol/L and addition of 5% DMSO, annealing temperature 50jC ! 43jC for 14 cycles and 40jC for 20 cycles. Sequencing was carried out using the ABI

Fig. 1. Structure of ERh gene. Position of htSNPs. Shaded boxes, exons; thin line, introns, promotor, 3V-UTR.

1937


Downloaded from clincancerres.aacrjournals.org on September 2, 2018. © 2006 American Association for Cancer Research.

Cancer Prevention

Table 1. Initially selected 37 SNPs for genotyping and construction of haplotypes dbSNP no.

Position

Exchange

rs2987983 rs1271572 rs1887994 rs1952586 rs1256028 rs1256029 rs1256030 rs1256031 rs960070 rs2776606 rs1256037 rs1256038 rs1273196 rs1256040 rs2026086 rs1256042 rs1256044 rs1256045 rs1256048 rs1256049 rs1256050 rs1977487 rs1256053 rs1256054 rs1256055 rs1952585 rs953592 rs2776608 rs1256060 rs944461 rs1256062 rs1256063 rs944045 rs944050 rs1256065 rs1152577 rs1152579

13950 12214 10908 9716 7446 5454 2533 3524 4524 5558 6402 8964 10198 11309 12348 15136 15676 19943 21423 25652 28264 31573 32748 33390 35530 37348 38954 40532 41653 44563 46385 47486 48257 9 -3 50771 52218 54616

T/C G/T G/T A/G A/G A/G C/T C/T C/G C/T C/T A/G C/T C/T C/T C/T G/T G/T G/T A/G C/T A/G G/T C/G A/G A/G A/G A/T A/G C/T A/G C/T A/G A/G A/C G/T A/G

Region

Promotor Promotor Intron 1 Intron 1 Intron 1 Intron 1 Intron 2 Intron 3 Intron 3 Intron 3 Intron 3 Intron 3 Intron 3 Intron 3 Intron 3 Intron 4 Intron 4 Intron 4 Intron 4 Exon 6 Intron 6 Intron 6 Intron 6 Exon 7 Intron 7 Intron 7 Intron 7 Intron 7 Intron 7 Intron 7 Intron 7 Intron 7 Intron 8 Intron/exon boundary 3V-UTR 3V-UTR 3V-UTR

Minor allele frequency in 94 subjects (%) 25.5 49.5 11.7 Assay failure Assay failure Assay failure 50.0 50.0 0 45.7 49.5 Assay failure 0 50.0 0 0 49.5 48.4 Assay failure 3.7 Assay failure Assay failure 3.7 0 3.7 3.2 Assay failure Assay failure 3.8 3.7 6.9 Assay failure 3.2 3.2 48.9 48.9 48.9

P

0.03 0.94

0.93

0.81

NOTE: htSNPs in boldface. Position is relative to the ATG start codon according to the National Center for Biotechnology Information genomic contig NT___026437.

Prism Big dye terminator kit v1.1/3.1 from Applied Biosystems and following the manufacturer’s instructions. When genotyping the whole study group, we placed two positive controls for each genotype and two negative controls on each plate. In addition, 29 blind duplicates were spread among the plates. Statistical analysis. Hardy-Weinberg Equilibrium tests for each sequence variant and pair-wise linkage disequilibrium tests for all sequence variants were done using a replication method as implemented in the GENETICS package (http://lib.stat.cmu.edu/R/CRAN/ index.html) for the R programming language. For each test, 10,000 permutations were done. Associations between genotypes and prostate cancer were assessed by the score test in unconditional logistic regression assuming a dominant genetic model. Genotype-specific risks were estimated as odds ratios with associated 95% confidence intervals by unconditional logistic regression. When testing for association and estimating odds ratios, the unconditional logistic regression was


stratified by each combination of age (5-year age groups) and geographic region to adjust for the matching conducted in collecting control subjects. Tests for association between haplotypes and prostate cancer risk were done using a score test developed by Schaid et al. (19), using the HAPLO.STAT program (http://www.mayo.edu/hsr/Sfunc.html) for the R programming language. This method, based on the generalized linear model framework, allows adjustment for possible confounding variables and provides both global tests and haplotype-specific tests. In these analyses, age and geographic region were adjusted for through indicator variables representing each combination of age category (5-year age groups) and geographic region (northern and central part of Sweden versus southeastern part of Sweden and the area of Stockholm). Haplotypes with estimated frequencies 0.05). The pair-wise linkage disequilibrium between these SNPs was 0.05 between rs2987983 and rs1256062 and 0.61 to 0.99 for the other SNP combinations. No genotyping error was detected among the duplicates, corresponding to an estimated error rate of 0.0%. The promoter SNP C/T 13950 (rs2987983) differed significantly in genotype frequency among cases and controls (P = 0.03). The genotype frequency of TC or CC was 47.6% in cases and 42.2% in controls. The difference was more pronounced among heterozygous carriers. The frequency for the C allele was 27% in cases and 24% in controls. For the other three htSNPs, no significant difference was found in genotype frequencies. Logistic regression analyses revealed a 23% increased risk of prostate cancer (odds ratios, 1.23; 95% confidence interval, 1.02-1.49) among men with the TC or CC compared with the genotype TT (adjusted for age and geographic region). Subgroup analysis based on localized or advanced cancer revealed an increased risk of 35% (odds ratios, 1.35; 95% confidence interval, 1.09-1.68) for being diagnosed with localized cancer (Table 2). Subgroup analysis by age (